Humanity is building supercomputers that can carry out a quintillion computations per second. However, these supercomputers still rely on the same principles from the earliest days of the computing revolution in the 1960s.
Quantum computing, which exploits the laws of quantum mechanics to create more efficient and powerful systems, is a promising alternative, but it is still in its infancy and faces many challenges. With the recent explosion of AI models, there’s a huge demand for computers that can process large sets of information. However, the current computing systems are quite inefficient and consume a lot of energy.
Now, researchers at the University of Pennsylvania have developed a new computer chip that uses light waves instead of electricity to perform the complex math essential to training AI. This could really improve the speed of data transfer and reduce the amount of electricity consumed, which would be fantastic for the environment.
The team has designed a silicon-photonic (SiPh) chip that brings together the pioneering research of Benjamin Franklin Medal Laureate and H. Nedwill Ramsey Professor Nader Engheta in manipulating materials at the nanoscale to perform mathematical computations using light with the SiPh platform. Silicon, a cheap and abundant element, is used to mass-produce computer chips.
The interaction of light waves with matter represents one possible avenue to develop computers that could surpass the limitations of today’s chips.
The researchers aimed to design a platform for performing vector-matrix multiplications, which is a core mathematical operation in the development of neural networks. They achieved this by making the silicon thinner in specific regions instead of using a silicon wafer of uniform height. These changes in height – without adding any other materials – allow them to control the propagation of light through the chip. The variations in height can be distributed to cause light to scatter in specific patterns, enabling it to perform mathematical calculations at the speed of light.
Firooz Aflatouni, Associate Professor in Electrical and Systems Engineering, who was involved in the research, said that this new design is already ready for commercial applications. It could potentially be adapted for use in graphics processing units (GPUs), which have become increasingly popular due to the widespread interest in developing new AI systems.
“They can adopt the Silicon Photonics platform as an add-on,” says Aflatouni, “and then you could speed up training and classification.”
The chip is said to be faster, consume less energy than existing designs, and offer privacy advantages. Because many computations can happen simultaneously, there is no need to store sensitive information in a computer’s working memory, making a future computer powered by this technology virtually unhackable.
“No one can hack into a non-existing memory to access your information,” added Aflatouni in a press release.
Journal reference:
- Vahid Nikkhah, Ali Pirmoradi, Farshid Ashtiani, Brian Edwards, Firooz Aflatouni, and Nader Engheta. Inverse-designed low-index-contrast structures on a silicon photonics platform for vector–matrix multiplication. Nature Photonics, 2024; DOI: 10.1038/s41566-024-01394-2